[BUG] PG restored cluster is always in Creating status due to Readiness probe failed

tianyue86 commented 2 months ago

Describe the bug

Kubernetes: v1.29.7-gke.1274000
KubeBlocks: 0.9.1-beta.25
kbcli: 0.9.1-beta.10

To Reproduce Steps to reproduce the behavior:

Create pg cluster kbcli cluster create postgres-icpzcz --termination-policy=DoNotTerminate --cluster-definition=postgresql --enable-all-logs=false --cluster-version=postgresql-14.8.0 --set cpu=100m,memory=0.5Gi,replicas=2,storage=3Gi --namespace default
Create backup kbcli cluster backup postgres-icpzcz --method wal-g --namespace default
Restore cluster kbcli cluster restore postgres-icpzcz-backup --backup backup-default-postgres-icpzcz-20240914124229 --namespace default

Check cluster status

tianyue@localhost kbcli % k get cluster -A | grep postgres
default     postgres-icpzcz          postgresql            postgresql-14.8.0     DoNotTerminate       Running    61m
default     postgres-icpzcz-backup   postgresql            postgresql-14.8.0     DoNotTerminate       **Creating**   38m

See error

tianyue@localhost kbcli % k describe cluster postgres-icpzcz-backup
Name:         postgres-icpzcz-backup
Namespace:    default
Labels:       clusterdefinition.kubeblocks.io/name=postgresql
          clusterversion.kubeblocks.io/name=postgresql-14.8.0
Annotations:  kubeblocks.io/ops-request: [{"name":"postgres-icpzcz-backup","type":"Restore"}]
          kubeblocks.io/reconcile: 2024-09-14T05:19:49.986719383Z
          kubeblocks.io/restore-from-backup:
            {"postgresql":{"connectionPassword":"EHhYeZrgFEC+x5rv7D+WRo9kZNvT2sIqM40QfqndWwQQIx94","doReadyRestoreAfterClusterRunning":"false","name":...
API Version:  apps.kubeblocks.io/v1alpha1
Kind:         Cluster
Metadata:
Creation Timestamp:  2024-09-14T04:44:04Z
Finalizers:
cluster.kubeblocks.io/finalizer
Generation:        1
Resource Version:  9186555
UID:               c84b3271-c513-450b-b3ef-55c9d37dd378
Spec:
Affinity:
Pod Anti Affinity:     Preferred
Tenancy:               SharedNode
Cluster Definition Ref:  postgresql
Cluster Version Ref:     postgresql-14.8.0
Component Specs:
Component Def Ref:  postgresql
Disable Exporter:   true
Enabled Logs:
  running
Name:      postgresql
Replicas:  2
Resources:
  Limits:
    Cpu:     100m
    Memory:  512Mi
  Requests:
    Cpu:               100m
    Memory:            512Mi
Service Account Name:  kb-postgres-icpzcz
Switch Policy:
  Type:  Noop
Volume Claim Templates:
  Name:  data
  Spec:
    Access Modes:
      ReadWriteOnce
    Resources:
      Requests:
        Storage:  3Gi
Resources:
Cpu:     0
Memory:  0
Storage:
Size:              0
Termination Policy:  DoNotTerminate
Status:
Cluster Def Generation:  2
Components:
Postgresql:
  Phase:       Creating
  Pods Ready:  false
Conditions:
Last Transition Time:  2024-09-14T04:44:04Z
Message:               The operator has started the provisioning of Cluster: postgres-icpzcz-backup
Observed Generation:   1
Reason:                PreCheckSucceed
Status:                True
Type:                  ProvisioningStarted
Last Transition Time:  2024-09-14T04:44:04Z
Message:               Successfully applied for resources
Observed Generation:   1
Reason:                ApplyResourcesSucceed
Status:                True
Type:                  ApplyResources
Last Transition Time:  2024-09-14T04:44:04Z
Message:               pods are not ready in Components: [postgresql], refer to related component message in Cluster.status.components
Reason:                ReplicasNotReady
Status:                False
Type:                  ReplicasReady
Last Transition Time:  2024-09-14T04:44:04Z
Message:               pods are unavailable in Components: [postgresql], refer to related component message in Cluster.status.components
Reason:                ComponentsNotReady
Status:                False
Type:                  Ready
Observed Generation:     1
Phase:                   Creating
Events:
Type     Reason                    Age                   From                  Message
----     ------                    ----                  ----                  -------
Normal   PreCheckSucceed           38m                   cluster-controller    The operator has started the provisioning of Cluster: postgres-icpzcz-backup
Normal   ApplyResourcesSucceed     38m                   cluster-controller    Successfully applied for resources
Normal   NeedWaiting               38m (x6 over 38m)     component-controller  waiting for restore "postgres-icpzcz-backup-postgresql-c84b3271-preparedata" successfully
Normal   ComponentPhaseTransition  38m                   cluster-controller    component is Creating
Warning  Unhealthy                 2m48s (x14 over 37m)  event-controller      Pod postgres-icpzcz-backup-postgresql-0: **Readiness probe failed**: 127.0.0.1:5432 - no response
Warning  Unhealthy                 2m41s (x12 over 36m)  event-controller      Pod postgres-icpzcz-backup-postgresql-1: **Readiness probe failed**: 127.0.0.1:5432 - no response

Expected behavior A clear and concise description of what you expected to happen.

Screenshots If applicable, add screenshots to help explain your problem.

Desktop (please complete the following information):

OS: [e.g. iOS]
Browser [e.g. chrome, safari]
Version [e.g. 22]

Additional context Add any other context about the problem here.

shanshanying commented 2 months ago

hi @tianyue86

to backup and restore a PG cluster using wal-g you should

config wal-g

kbcli cluster backup <clusterName> --method config-wal-g

update parameters using ops

apiVersion: apps.kubeblocks.io/v1alpha1
kind: OpsRequest
metadata:
generateName: pg-cluster-reconfiguring-
spec:
clusterRef: <clusterName>
reconfigure:
componentName: postgresql
configurations:
  - keys:
      - key: postgresql.conf
        parameters:
          - key: archive_command
            value: "'envdir /home/postgres/pgdata/wal-g/env /home/postgres/pgdata/wal-g/wal-g wal-push %p'"
    name: postgresql-configuration
type: Reconfiguring

backup cluster

kbcli cluster backup <clusterName> --method wal-g

restore

kbcli cluster restore <clusterName> --backup <backupName>

shanshanying commented 2 months ago

apecloud / kubeblocks

[BUG] PG restored cluster is always in Creating status due to Readiness probe failed #8152